Classification of Dental Caries Using the Modified K-Nearest Neighbor Algorithm

Abstract

Dental caries, commonly known as tooth cavities, is a disease where bacteria damage the structure of tooth tissues such as enamel, dentin, and cementum. The primary cause of dental caries is the demineralization of tooth surfaces caused by organic acids from sugary foods. If dental caries is not promptly treated or checked from the beginning, the damage can worsen to the point where the tooth must be extracted. To facilitate identifying the severity of caries, a dental caries classification system was developed using the MKNN (Modified K-Nearest Neighbor) algorithm. The MKNN method is an enhancement of the KNN method, with the main differences being in the calculation of training data validity and the weight voting process. In this study, there are three different classes of dental caries and six symptoms or variables. The stages of the MKNN method used are: distance calculation using Euclidean distance, testing the validity of training data, determining k based on distance calculation, and weight voting calculation in KNN. The test results show that the k value, the number of training data, and the number of test data affect the classification results. The classification results from the test using 20 training data, 10 test data, and k=3 are as follows: 1 patient classified with superficial caries, 5 patients with media caries, and 3 patients with profunda caries. The diagnosis produced by the application is consistent with the expert (doctor) diagnosis.